Priority Poker Delta
Purpose: Rank pokerplan facilitation and bias-control contract. Planning Poker and Fibonacci scales are model-known.
Protocol
- Define the dimension being voted on; never mix value, effort, risk, and confidence into one unexplained card.
- Select low/mid/high reference items and calibrate them before novel items.
- Vote privately and reveal simultaneously.
- Treat
?as missing context, not zero. - When votes are dispersed, highest and lowest voters explain first; re-vote at most three rounds.
- Persistent disagreement is an input to Magi, not a reason to force consensus.
Default dispersion gate: re-discuss when the spread crosses more than two Fibonacci steps. Teams may choose another gate, but must record it before voting.
Required Output
| Item | Dimension | Final votes | Aggregate | Dispersion | Rounds | Confidence | Key disagreement |
|---|
Record anchors, participants or represented roles, missing-context cards, and any facilitator override. Feed the resulting component scores into ICE/RICE/WSJF only after preserving their units.
Reject
- Sequential reveal or leader-first discussion.
- Uncalibrated sessions whose scores are compared across teams or dates.
- Averaging
?cards into the result. - More rounds used to wear down dissent.
Handoff
- Formula scoring -> Rank recipe matching the chosen dimensions.
- Structural disagreement -> Magi.
- Top item decomposition -> Sherpa.
- Missing feature substance -> Spark.